What we read
Toe Tag reads memecoins on Solana from public on-chain and market data. Our daily sweep looks at tokens that migrated from their launch curve in the last 14 days, and keeps the ones with a trusted trading pair and at least $40,000 in liquidity. That is the "first cut" in our numbers.
When the Check opens, it will also tell you when a token is outside what we can read yet: still on its launch curve, on an unsupported launchpad, or without a trusted pool. A reading never pretends to cover what it doesn't.
What the filter looks at
Every token that passes the first cut goes through the same set of structural checks, in the same order. Each check looks for a way the game can be set up against the buyer:
- Token controls. Whether someone can still mint more supply, freeze wallets, or uses token extensions that change how transfers work.
- The creator at launch. Whether the creator bought or received a large share at creation, and whether the creator wallet is handing out supply to other wallets.
- Token factories. Whether the same creator keeps launching token after token.
- The first buyers. Bundled launches (many wallets buying in the same slot), wallets funded by a common source, and how concentrated the early supply is.
- Holders. How much of the supply the top 10 holders own, with liquidity pools excluded.
- Liquidity and the pair. Whether we're looking at the trusted pool, recognized by address, and how much real liquidity it has.
- Stale prices. Whether almost all the supply is back in the pool, a sign that a chart can look alive when nobody is trading.
- The address itself. Whether the address matches the source you got it from, so a copycat with the same name doesn't pass as the real one.
We publish what each check looks at and what it means. We don't publish exact thresholds: a public number is a target for the people who launch these tokens.
The four states
| State | What it means |
|---|---|
| Passed | It passed the structural checks. It does not mean "buy" and it does not mean "will survive". Most tokens die, including many that pass. |
| Discarded | It failed at least one check. The reading says which one, under Tagged for. |
| Review | Something needs a closer look before it can be called either way. |
| Incomplete | The reading couldn't be finished: a source didn't respond, or the token is outside what we can read. |
Inside a reading, a single data point we couldn't verify is marked Missing. Never "none", never zero, never left blank.
How we count a death
We fixed this definition on November 1, 2026, before looking at the numbers.
Dead: 7 days after our reading, the token's final price is 90% or more below the price at our reading.
- Prices come from GeckoTerminal, on the pool we registered when we read the token. The reading price is the one we stored; the final price is the close of the last full hourly candle at 7 days.
- No trades at all in 7 days is not counted as dead or alive. It's reported apart, as "total silence".
- No trusted pool when we read it, or a missing pool: left out of the count, and reported as left out.
- One token, one count: each token counts once, with its first reading.
- Series: from October 10, 2026. Our first census includes readings from the period when we were testing the sweep; the prices and the deaths in it are real.
Discarded is not dead
Some tokens our filter discards keep trading, sometimes with market caps in the millions. A discarded token gets a Tagged for line with the reason. Cause of Death and The Obits are only for tokens that meet the definition above.
No pulse
When the Check opens, we'll also flag tokens with No pulse: tokens that look alive on a chart but aren't. It fires on one of two facts: almost all of today's supply is back in the main pool, or there were zero trades in the main pool in the last 24 hours. Seven days later we publish how many of those tokens died or went fully silent. It's a report of facts, not a prediction and not a signal.
When we publish a number
- Fewer than 30 cases: we show the count, never a rate.
- From 30 to 99 cases: still no published rate.
- 100 cases or more: we publish the rate, always with the number of cases and a 95% confidence interval.
- If a threshold was set using the same tokens we're reporting on, we say so next to the number.
A mortality rate describes how past tokens with a similar pattern ended. It is history, not a prediction about the token in front of you.
What we publish, and what we don't
- We publish discards with their reason, case files of measured deaths, reading manuals, and our full results, including the ones that make us look bad. We never pick cases by hand.
- We never publish a live list of tokens that passed. A list like that works as a buy list, and it gets gamed.
- We don't score upside, and we don't say what to buy, when, how much, or at what price.
Sources and credit
On-chain data from Solana. Market data and price history from GeckoTerminal (Powered by CoinGecko) and DexScreener. Charts on this site are our own, built from our own records. Each figure carries its sample, period and source.
Last updated: November 1, 2026